JPH0439609B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a control device for a chassis dynamometer.

[Conventional technology]

A DC chassis dynamometer absorbs the force F _veh generated by a car using electrical force, simulates the state of a car driving on a road, and can perform various indoor driving tests if possible. This is a simulator. As a control method for a chassis dynamometer, a torque control method that combines feedforward control and feedback control, as shown in FIG. 1, has conventionally been used. In the figure, 1 is a torque sensor, 2 is a flywheel, and 3
4 is a roller on which the driving wheels of the test vehicle are placed, 4 is a speed sensor, and 5 is a roller shaft.

In this configuration, the force F _veh (t) generated by the vehicle at time t is measured by the torque sensor 1 as the output F _TT (t), and the speed V(t) is determined by the speed sensor 4. Ru. These measured values are then input to the feedforward control circuit 6 and the error function calculation circuit 7. The feed forward control circuit 6 uses the above measured value at time t to perform the next step (t+ after ΔT time).
The force F _PAU that the dynamometer should absorb at ΔT)
(t+ΔT) is calculated based on the following equation.

F _PAU (t+ΔT)=I _n +I _r /IRL(V)+I _e /IF _TT (t
)＋I^ _n dv／dt＋L _n (V)…(1) Where, I^m; Set weight of flywheel (Kg) I^ _r ; Set weight of roller in dynamo (Kg) I^; Equivalent to car Inertia weight (setting) (Kg) I _e ;=I^-(I^ _n + I^ _r ) RV (V); Running resistance (=A+BV+CV ^x ) _Ln (V): Mechanical loss of the flywheel.

On the other hand, an error function calculation circuit 7 calculates an error function. Here, the error function is the actual predicted output value of the dynamometer, F′ _PAU , including the inertial force of the roller.
(t+ΔT) and the target value F _TT (t)+L _n (V), and is given by the following equation.

∫ ^t ₀ ε(t)dt＝∫ ^t ₀ {F _TT (t)+L _n (V)−F′ _PAU
(t+
△T)}dt =∫ ^t ₀ {F _TT (t)+L _n (V)−RL(V)}dt−(I^
-I^m).V (2) This error signal is input to the adder 9 through the PI control circuit 8, and the feedforward-controlled signal F _PAU (t+ΔT) is corrected. The modified signal is expressed as:

F _PC (t)=F _PAU (t+ΔT)+{∫ ^t ₀ ε(t)dt}・(
K _P )+K _I /S)...(3) Here, K _P : Coefficient of proportional term of PI control K _I : Coefficient of integral term of PI control S: Laplace operator.

This correction signal P _PC (t) controls the power converter 10 and causes an exciting current to flow through the DC motor/dynamo 11 as a dynamometer. This controls the DC motor/dynamo 11 to absorb the force exerted by the vehicle.

However, since the conventional control method described above is mainly based on feed forward control, it has the following drawbacks. That is, since the feed-forward control method is a method of predicting the force to be absorbed after a time ΔT at time t, deviations in prediction often occur. If the prediction deviation is within a small range, it can be corrected by the error function, but if it exceeds the range that can be corrected by the error function, the prediction deviation becomes large, causing the control system to oscillate. This situation occurs, for example, when the weight of the car is light;
The test car goes out of control.

[Purpose of the invention]

Therefore, the present invention aims to eliminate the above-mentioned drawbacks by adopting a new method in place of the above-mentioned system that combines feedback control and feedback control.

[Structure of the invention]

In order to achieve the above object, in the present invention, a torque sensor, a speed sensor, and a flywheel are provided on the shaft directly connected to the roller dynamometer on which the drive wheels of the test vehicle are mounted, and a control signal from the power converter is provided. In a control device for a chassis dynamometer in which the dynamometer absorbs the force output by the vehicle, {F _TT (t) + L _n (V) - F' _PAU (t + ΔT)} ² (However, F _TT (t): Torque sensor output, L _n (V): Mechanical loss of flywheel, F _TT (t) + L _n (V): Set target value F' _PAU (t + ΔT): Actual output of dynamometer The search direction vector P(t) is calculated using a calculation circuit that calculates the gradient of the evaluation function J expressed by
An arithmetic circuit that determines F _PAU (t) + αR (t) (where P (t) = βP (t −ΔT)−▽J, F _PAU (t) is the step signal in the power converter,
α, β are constants) is converted to the next step signal F _PAU (t+
ΔT).

That is, as the evaluation function, choose J = {F _TT (t) + L _n (V) - F _PAU (t + ΔT)} ² ... (4), and use the following conjugate gradient method algorithm F _PAU (t + ΔT) = F _PAU ( t)+α・P(t) …(5) P(t)=β・P(t-ΔT)−Δ・J=β・P(t
−ΔT) −{F _TT (t) + L _n (V) − RV (V) − I^ − I^ _n ) dv/dt
} ...(6), it is possible to adapt to system fluctuations such as acceleration/deceleration and make sequential corrections, thereby achieving quick response and stabilization of the control system. In the above equations (5) and (6), α and β are constants with small values.

〔Example〕

FIG. 2 shows an embodiment of the present invention, in which the first
Components and elements that are the same as those used in the control device shown in the figures are given the same numbers and their explanations will be omitted. 12 is a circuit that differentiates the speed signal V with time t; 13 is an evaluation based on the following formula using the output F _TT (t) of the torque sensor 1, the output V of the speed sensor 4, and the output dv/dt of the differentiating circuit 12. This is a circuit that calculates the gradient Δ·J of a function.

Δ・J=F _TT (t) + L _n (V) − RL (V) − (I^−I^ _n )
dv/dt...(7) 14 is a circuit that determines the search direction vector {P(t)} by the conjugate gradient method, which includes a β multiplier 15 and a β
a subtraction unit 16 that subtracts the signal Δ·J of the Δ·J calculation circuit 13 from the output signal β·P(t) of the multiplier 15;
Based on the signal from the subtractor 16, {P
(t)}.
The search direction vector {P(t)} is expressed by the above equation (6).

18 is a drive signal F _PAU (t+
This is a circuit that sequentially corrects the search vector {P
(t)} by α, an α multiplier 19, and an adder 20
and a drive signal F _PAU (t+ΔT) generating section 21. Here, drive signal F _PAU (t+ΔT)
is expressed by equation (5).

The drive signal F _PAU (t+
ΔT) is applied to the power converter 10, causing an excitation current to flow through the DC motor/dynamo 11. This allows the dynamometer to measure the F _veh (t) generated by the test vehicle.
controlled to absorb.

It is clear that the feedback circuit from the Δ/J calculation circuit 13 to the sequential correction circuit 18 can also be realized by a digital circuit or software processing by a microcomputer or the like.

Since the control device for a chassis dynamometer according to the present invention is constructed as described above, it has the following effects.

Since the conjugate gradient method is used instead of the feedforward control method, there is no possibility that the control system would oscillate under certain conditions, as was the case in the past.
A stable control system can be obtained.

Using the conjugate gradient method, which finds the minimum point by repeating the search along the conjugate direction,
Since optimal control is performed to minimize the evaluation function given by equation (4), the control system can respond more quickly.

[Brief explanation of drawings]

FIG. 1 is a diagram showing a conventional control device for a chassis dynamometer, and FIG. 2 is a diagram showing a control device for a chassis dynamometer as an embodiment of the present invention. 1...Torque sensor, 2...Flywheel,
3...Roller, 4...Speed sensor, 5...Shaft, 10...Power converter, 11...Dynamometer, 13...Arithmetic circuit, 14...Circuit for calculating the search direction vector, 18...Sequential Modified arithmetic circuit.

Claims (1)

[Claims] 1. A torque sensor, a speed sensor, and a flywheel are provided on a shaft that directly connects the dynamometer to the roller on which the drive wheels of the test vehicle are mounted, and the dynamometer is controlled by a control signal from a power converter. In a control device for a chassis dynamometer that absorbs the force output from the vehicle, {F _TT (t) + L _n (V) - F' _PAU (t + △T)} ² (However, F _TT (t): Torque sensor output, L _n (V): Mechanical loss of flywheel, F _TT (t) + L _n (V): Set target value F' _PAU (t + ΔT): Expected actual output of dynamometer The search direction vector P(t) is calculated using a calculation circuit that calculates the gradient of the evaluation function J expressed as
An arithmetic circuit that determines F _PAU (t) + αP (t) (where P (t) = βP (t −ΔT)−▽J, F _PAU (t) is the step signal in the power converter,
α, β are constants) is converted to the next step signal F _PAU (t+
∆T).